|
--- |
|
license: apache-2.0 |
|
base_model: t5-base |
|
tags: |
|
- generated_from_trainer |
|
datasets: |
|
- eli5 |
|
metrics: |
|
- rouge |
|
model-index: |
|
- name: eli5 |
|
results: |
|
- task: |
|
name: Sequence-to-sequence Language Modeling |
|
type: text2text-generation |
|
dataset: |
|
name: eli5 |
|
type: eli5 |
|
config: LFQA_reddit |
|
split: validation_eli5 |
|
args: LFQA_reddit |
|
metrics: |
|
- name: Rouge1 |
|
type: rouge |
|
value: 14.6325 |
|
--- |
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
# eli5 |
|
|
|
This model is a fine-tuned version of [t5-base](https://huggingface.co/t5-base) on the eli5 dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 2.2569 |
|
- Rouge1: 14.6325 |
|
- Rouge2: 2.3714 |
|
- Rougel: 11.2941 |
|
- Rougelsum: 13.2006 |
|
- Gen Len: 18.9911 |
|
|
|
## Model description |
|
|
|
More information needed |
|
|
|
## Intended uses & limitations |
|
|
|
More information needed |
|
|
|
## Training and evaluation data |
|
|
|
More information needed |
|
|
|
## Training procedure |
|
|
|
### Training hyperparameters |
|
|
|
The following hyperparameters were used during training: |
|
- learning_rate: 2e-05 |
|
- train_batch_size: 8 |
|
- eval_batch_size: 8 |
|
- seed: 42 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- num_epochs: 2 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Rouge1 | Rouge2 | Rougel | Rougelsum | Gen Len | |
|
|:-------------:|:-----:|:-----:|:---------------:|:-------:|:------:|:-------:|:---------:|:-------:| |
|
| 2.4057 | 1.0 | 34080 | 2.2708 | 14.6356 | 2.3501 | 11.3428 | 13.213 | 18.9946 | |
|
| 2.3943 | 2.0 | 68160 | 2.2569 | 14.6325 | 2.3714 | 11.2941 | 13.2006 | 18.9911 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.34.0 |
|
- Pytorch 2.1.0 |
|
- Datasets 2.14.5 |
|
- Tokenizers 0.14.1 |
|
|
